pakawadeep/mt5-base-finetuned-ctfl-backtranslation_7k

This model is a fine-tuned version of google/mt5-base on an unknown dataset. It achieves the following results on the evaluation set:

• Train Loss: 5.0781
• Validation Loss: 4.5021
• Train Bleu: 0.0
• Train Gen Len: 21.0
• Epoch: 16

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• optimizer: {'name': 'AdamWeightDecay', 'learning_rate': 2e-05, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-07, 'amsgrad': False, 'weight_decay_rate': 0.01}
• training_precision: float32

Training results

Train Loss	Validation Loss	Train Bleu	Train Gen Len	Epoch
16.3214	7.9429	0.0008	127.0	0
9.3082	7.3301	0.0	127.0	1
8.1593	7.1451	0.0	127.0	2
7.7440	7.0581	0.0	8.0	3
7.5624	7.0120	0.0	3.0	4
7.4569	6.9663	0.0	3.0	5
7.3873	6.9154	0.0001	114.0	6
7.3206	6.8373	0.0001	127.0	7
7.2175	6.7063	0.0001	127.0	8
7.0942	6.4250	0.0003	127.0	9
6.8748	6.0066	0.0003	127.0	10
6.5383	5.5921	0.0003	127.0	11
6.2094	5.2557	0.0003	127.0	12
5.8157	4.9725	0.0003	127.0	13
5.5098	4.7642	0.0003	127.0	14
5.2683	4.6171	0.0017	127.0	15
5.0781	4.5021	0.0	21.0	16

Framework versions

• Transformers 4.44.2
• TensorFlow 2.17.0
• Datasets 3.0.1
• Tokenizers 0.19.1